Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I have definitely noticed that ChatGPT is atrocious at writing Polars code (which was written recently and has a changing API) while being good at Pandas. I figure this will mostly resolve when the standard reasoning models incorporate web search through API documentation + trial and error code compilation into their chain of thought.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: